Verification · Code review · LLM application security review
Runtime input and output scanning
Scan prompts and responses in production for injection and leakage.
How to build it
- Place a scanner in front of the prompt path and behind the response path.
- Block or redact the content the scanner flags, and log the flag without the content.
- Track the rate of flagged inputs and outputs per route.
Signals in a repository
- Scanner calls that wrap the request and response functions in the service.
- A guardrails or rails config directory in the repository.
- Logs or metrics that count scanner flags per route.
In practice
No scanned repository runs a tool for this check. The steps above are the only guidance.